Related Experiment Video
Updated: May 12, 2026
07:41
Behavioral Assessment of Visual Function via Optomotor Response and Cognitive Function via Y-Maze in Diabetic Rats
Published on: October 23, 2020
6.7K
Enhancing diabetic retinopathy diagnosis and grading: a retrospective study on AI-assisted decision making and cost
Xieyang Xu1,2, Jiaying Zhang1,2, Xuefei Song1,2
1Department of Ophthalmology, Shanghai Jiao Tong University School of Medicine Affiliated Ninth People's Hospital, Shanghai, China.
The British Journal of Ophthalmology
|October 20, 2025
Summary
Artificial intelligence (AI) significantly improved diabetic retinopathy (DR) diagnosis accuracy when used as an assistant tool by physicians. This AI assistance enhances early detection and treatment of DR, despite associated costs.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Diabetic retinopathy (DR) is a leading cause of blindness.
- Current artificial intelligence (AI) research for DR primarily compares AI to human performance.
- Limited research explores AI's role as an assistive tool in DR diagnosis and grading.
Purpose of the Study:
- To evaluate the impact of AI-assisted decision-making on DR diagnosis and grading.
- To assess AI's effectiveness using both color fundus photographs (CFP) and ultra-widefield fundus (UWF) images.
Main Methods:
- 224 retinal images were analyzed by 21 ophthalmologists and primary care physicians (PCPs).
- Participants diagnosed and graded DR with and without AI assistance.
- Diagnosis accuracy was compared to a gold standard; incremental costs and accuracy improvements were assessed using generalized estimating equations (GEE) models.
Main Results:
- AI assistance significantly improved DR diagnosis accuracy for both CFP and UWF images across all physician groups.
- For CFP, accuracy increased by 5-7% with AI assistance.
- For UWF, accuracy improved by approximately 6% with AI assistance.
Conclusions:
- AI assistance holds significant potential for enhancing DR diagnosis and grading accuracy.
- AI enables ophthalmologists to achieve superior diagnostic performance, aiding in earlier DR detection and treatment.
- The benefits of AI assistance in improving diagnostic accuracy outweigh the associated costs.
